An Overview of the Theory and Application of Reinforcement Learning

As an important branch of machine learning research, reinforcement learning can obtain strategy improvement through the interaction of trial and environment and is widely applied in various fields. Research on reinforcement learning theory is helpful for subsequent multidisciplinary projects. However, related researches are still relatively scattered. Thus, this paper introduces the theory and application of reinforcement learning by literature analysis and comprehensively introduces the main algorithms theory of reinforcement learning, including temporal difference learning, Q-learning and Sarsa learning, as well as their combination and effect comparison. Besides, this paper also summarizes the current applications of reinforcement learning that have received more attention, namely control systems, autonomous driving, and robots. Moreover, the current research issues and future work directions of reinforcement learning are discussed. Overall, this article concludes that the current basic algorithm construction of reinforcement learning is relatively complete, but the method of combining multiple algorithms remains to be discussed.